A data processing method and system for computer-aided performance optimization design of building energy systems based on time-of-use carbon emission factors.
By using a multi-output MLP proxy model and a multi-objective particle swarm optimization algorithm, combined with time-of-use electricity carbon emission factors, the problem of insufficient low-carbon assessment and the unification of seasonal variables for cooling and heating in building energy systems is solved, enabling the rapid and effective generation of optimized design schemes.
Patent Information
- Application Number
- CN202611339567.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2026-09-01
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2046-09-01
AI Technical Summary
Existing proxy optimization methods for building energy systems lack low-carbon assessment, the system COP is affected by the energy supply, the variables of the cooling season and the heating season are difficult to express uniformly, the multi-objective optimization results are difficult to make quick engineering decisions, and the number of detailed simulation calls is high.
A multi-output MLP proxy model is adopted to replace detailed simulation software. Combining multi-objective particle swarm optimization algorithm and multi-scenario weight decision, the system COP is corrected through effective energy supply. The time-sharing power carbon emission factor is introduced to calculate the operating carbon emission index, generate Pareto front solution set that meets the constraints, and output the optimized design scheme.
It reduces the frequency of calling detailed simulation software, avoids artificially high efficiency caused by over-cooling or over-heating, reflects the differences in carbon intensity of electricity consumption at different times, and quickly generates optimized design schemes that are easy to use in engineering applications.
Smart Images

Figure CN122863713B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of computer-aided building performance design technology, and more specifically, to a data processing method and system for computer-aided building energy system performance optimization design oriented towards time-of-use carbon emission factors. Background Technology
[0002] Building energy systems are a crucial component influencing the energy consumption and carbon emissions of public buildings. Hybrid ground-source heat pump systems, chiller systems, air-source heat pump systems, and their coupled systems typically include multiple heat source and cooling source devices, water pumps, cooling towers, and terminal heat exchange units. Their operational performance is influenced by factors such as building heating and cooling loads, outdoor weather conditions, operating hours, equipment start-up and shutdown status, partial load rate, water pump flow rate ratio, supply and return water temperature difference, and regional or time-of-use electricity carbon emission factors.
[0003] In the performance design, retrofit design, operation scheme design, and operation parameter optimization of building energy systems, detailed simulation models can accurately describe the system's thermal and energy response. However, in multi-time period, multi-variable, and multi-objective optimization scenarios, calling detailed simulation models for performance evaluation of each candidate scheme leads to excessively long computation times, making it difficult to support rapid scheme comparison, rolling optimization, and engineering decision support. Existing surrogate model optimization methods typically use machine learning models trained on simulation samples as fast evaluation functions and combine them with multi-objective optimization algorithms to output Pareto solutions. However, ordinary surrogate optimization methods often use system energy consumption or system power as a single or primary evaluation indicator, easily ignoring the differences in electricity carbon emission factors at different times and failing to reflect the impact of low-carbon and high-carbon electricity periods on the optimization results of building energy system operation parameters.
[0004] Meanwhile, in the evaluation of the operation optimization of some building energy systems, if the ratio of the total cooling or heating capacity to the system power is directly used as the efficiency indicator, when the optimization algorithm generates over-cooling or over-heating schemes, the portion exceeding the building's cooling and heating load demand does not form an effective service, but may still be included in the energy supply, thus causing the system efficiency evaluation to be distorted and inducing the optimization algorithm to select a scheme with higher apparent efficiency but insufficient actual load matching.
[0005] Furthermore, the equipment configuration, operating equipment combination, and operating parameter variables of building energy systems may differ between the cooling and heating seasons. For example, ground source heat pumps and chillers may be used in combination during the cooling season, while ground source heat pumps and air source heat pumps may be used in combination during the heating season. The proportion of pump flow rates, equipment start-up and shutdown states, and operating characteristics involved in optimization also differ under different seasonal modes. Without a unified seasonal mode, equipment configuration, and equipment state masking encoding method, it is difficult to maintain consistency in surrogate model training, optimization variable definition, candidate scheme evaluation, and hourly operating parameter sequence output across different seasonal modes.
[0006] Therefore, there is an urgent need for a data processing method for computer-aided performance optimization design of building energy systems. This method should reduce the number of detailed simulation calls while taking into account the energy consumption, carbon emissions, effective energy supply efficiency, fluctuations in operating parameters, and the feasibility of engineering constraints of building energy systems. This would improve the low-carbon nature, interpretability, feasibility, and engineering applicability of energy-saving and carbon-reducing operating parameter optimization design schemes for building energy systems. Summary of the Invention
[0007] The purpose of this invention is to provide a computer-aided building energy system performance optimization design data processing method and system oriented towards time-of-use carbon emission factors, in order to solve the problems of insufficient low-carbon evaluation, system COP being affected by excessive energy supply, difficulty in uniformly expressing variables for cooling and heating seasons, difficulty in quickly outputting multi-objective optimization results for engineering decisions, and high number of detailed simulation calls in existing proxy optimization methods for building energy systems.
[0008] This invention is achieved through the following technical solution: A data processing method for computer-aided performance optimization design of building energy systems based on time-of-use carbon emission factors includes: Obtain the target parameters, and after preprocessing some of the target parameters, form a building energy system performance prediction input vector; A multi-output MLP proxy model is constructed and trained based on simulation model output data or actual operation monitoring data. The multi-output MLP proxy model predicts the cooling or heating capacity, total power and supply and return water temperature difference of the system according to the building energy system performance prediction input vector. The effective energy supply is determined based on the matching relationship between the cooling or heating capacity of the system and the building's cooling and heating load, and the effective COP of the system is calculated based on the effective energy supply and the total power of the system. Using the optimization sequence of operating parameter variables as candidate solutions, a constraint-aware multi-objective particle swarm optimization model (MOPSO) is constructed with the optimization objectives of minimizing system operating energy consumption, minimizing operating carbon emission indicators, maximizing system effective COP, and minimizing fluctuations in operating parameter variables. The multi-output MLP proxy model is embedded as a fast performance evaluation function into the constraint-aware multi-objective particle swarm optimization algorithm MOPSO to generate a Pareto front solution set that satisfies the constraints or has been processed for constraint violation. Based on the multi-scenario weighted TOPSIS, a compromise optimal solution is determined from the Pareto front solution set, and the energy-saving and carbon-reduction performance optimization design scheme and hourly operation parameter optimization sequence of the building energy system are output.
[0009] Secondly, the present invention also provides a computer-aided building energy system performance optimization design data processing system oriented towards time-of-use carbon emission factors, for executing the aforementioned computer-aided building energy system performance optimization design data processing method oriented towards time-of-use carbon emission factors, comprising: The building energy system design data acquisition module is configured to acquire target parameters and preprocess some of the target parameters to form a building energy system performance prediction input vector. Seasonal Equipment Configuration Unified Coding Module: This module generates equipment status masks and operating parameter mapping matrices based on seasonal patterns, equipment configurations, and equipment start-up and shutdown states. It maps different numbers and meanings of operating parameter variables for cooling and heating seasons into a fixed-dimensional unified operating parameter variable vector, and forms a building energy system performance prediction input vector and a low-carbon evaluation input vector. The multi-output MLP proxy model training and performance prediction module is configured to build and train a multi-output MLP proxy model based on simulation model output data or actual operation monitoring data. The multi-output MLP proxy model predicts the cooling or heating capacity, total power and supply and return water temperature difference of the system according to the building energy system performance prediction input vector. The effective energy supply and time-of-use carbon emission index calculation module is configured to determine the effective energy supply based on the matching relationship between the system's cooling or heating capacity and the building's cooling and heating load, and to calculate the system's effective COP based on the effective energy supply and the system's total power. The constraint violation handling and multi-objective optimization module is configured to use the optimization sequence of operating parameter variables as candidate solutions to construct a constraint-aware multi-objective particle swarm optimization (MOPSO) model with the optimization objectives of minimizing system operating energy consumption, minimizing operating carbon emission indicators, maximizing system effective COP, and minimizing fluctuations of operating parameter variables. The multi-scenario TOPSIS decision ranking and solution output module is configured to embed the multi-output MLP proxy model as a fast performance evaluation function into the constraint-aware multi-objective particle swarm optimization algorithm MOPSO, generate a Pareto front solution set that satisfies the constraints or has been processed for constraint violation, determine the compromise optimal solution from the Pareto front solution set based on the multi-scenario weighted TOPSIS, and output the energy-saving and carbon-reduction performance optimization design scheme of the building energy system and the hourly operation parameter optimization sequence.
[0010] The technical solution of the present invention has at least the following advantages and beneficial effects: The method provided by this invention replaces detailed simulation software with a multi-output MLP proxy model, reducing frequent calls to detailed simulation software during the optimization process and shortening the evaluation time for multi-objective optimization. It avoids artificially inflated efficiency caused by over-cooling or over-heating by effectively correcting the system's COP evaluation through energy supply correction. It introduces regional or time-of-use electricity carbon emission factors to calculate operational carbon emission indicators and constructs carbon time-series weights based on these factors, reflecting the impact of differences in electricity carbon intensity at different times on the optimization design results of building energy system operating parameters. Through constraint-aware multi-objective optimization and multi-scenario weight ranking, it outputs a compromise optimal solution, enabling the rapid generation of corresponding performance optimization design schemes that are easy for engineering applications based on priorities such as energy saving, low carbon, efficiency, stability, or equilibrium. Attached Figure Description
[0011] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.
[0012] Figure 1 This is a schematic diagram of the process of the present invention. Detailed Implementation
[0013] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.
[0014] Please refer to Figure 1 This invention provides a computer-aided building energy system performance optimization design data processing method based on time-of-use carbon emission factors, comprising: S1: Obtain the target parameters, and after preprocessing some of the target parameters, form a building energy system performance prediction input vector; The target parameters include the building's heating and cooling load, meteorological parameters, time characteristics, seasonal patterns, equipment configuration, equipment start-up and shutdown status, operating conditions, candidate operating parameter variables, and regional or time-of-use electricity carbon emission factors within the set optimization time domain.
[0015] Based on seasonal patterns, equipment configuration, and equipment start-up and shutdown status, an equipment status mask and an operating parameter mapping matrix are generated. Different numbers and meanings of operating parameter variables in the cooling season and heating season are mapped into a unified operating parameter variable vector of fixed dimensions. The unified operating parameter variable vector, building cooling and heating load, meteorological parameters, time characteristics, seasonal patterns, equipment configuration, equipment start-up and shutdown status, and equipment operating condition characteristics form the building energy system performance prediction input vector.
[0016] S2: Construct and train a multi-output MLP proxy model based on simulation model output data or actual operation monitoring data. The multi-output MLP proxy model predicts the system's cooling or heating capacity, total system power, and supply and return water temperature difference based on the building energy system performance prediction input vector. It determines the effective energy supply based on the matching relationship between the system's cooling or heating capacity and the building's cooling and heating loads, and calculates the system's effective COP based on the effective energy supply and total system power.
[0017] A multi-output MLP proxy model is constructed and trained based on simulation model output data or actual operation monitoring data. The multi-output MLP proxy model synchronously predicts the system's cooling or heating capacity, total system power, and supply and return water temperature difference according to the building energy system performance prediction input vector. The effective energy supply is determined based on the matching relationship between the system's cooling or heating capacity and the building's cooling and heating load. The effective COP of the system is calculated based on the effective energy supply and the total system power. When the system's cooling or heating capacity exceeds the building's cooling and heating load for the corresponding period, the excess supply exceeding the building's cooling and heating load is not included in the effective energy supply.
[0018] S3: When using time-of-use electricity carbon emission factor, construct carbon time series weights based on the ratio of time-of-use electricity carbon emission factor to the optimized time-domain average carbon emission factor or the normalization result, and calculate the operating carbon emissions, carbon emission intensity and power weighted index of high carbon period based on the total system power, control time step, time-of-use electricity carbon emission factor and carbon time series weights.
[0019] S4: Using the optimization sequence of operating parameter variables as candidate solutions, construct a constraint-aware multi-objective particle swarm optimization (MOPSO) model with the optimization objectives of minimizing system operating energy consumption, minimizing operating carbon emission indicators, maximizing system effective COP, and minimizing fluctuations in operating parameter variables. During the multi-objective optimization iteration process, the constraint violation degree of the candidate solution relative to the load satisfaction rate, supply and return water temperature difference, equipment start-up and shutdown logic, and operating parameter engineering constraints is calculated; when the constraint violation degree is greater than zero, variable pruning, feasible region projection, penalty function penalty, and / or non-dominated solution external file priority reduction processing are performed on the candidate solution. S5: The multi-output MLP proxy model is embedded as a fast performance evaluation function into the constraint-aware multi-objective particle swarm optimization algorithm MOPSO to generate a Pareto front solution set that satisfies the constraints or has been processed for constraint violation. Based on the multi-scenario weight TOPSIS, the compromise optimal solution is determined from the Pareto front solution set, and the energy-saving and carbon-reduction performance optimization design scheme of the building energy system and the hourly operation parameter optimization sequence are output.
[0020] In one exemplary embodiment of the present invention, the unified encoding of seasonal configurations includes: during the t-th control period, according to the seasonal pattern... and equipment configuration Determine the original operating parameter variable vector under the current seasonal configuration. The dimension of the original running parameter variable vector is denoted as ; According to seasonal patterns and equipment configuration Determine the runtime parameter mapping matrix The dimension of the runtime parameter mapping matrix is D× , where D is the dimension of the fixed-dimensional uniform runtime parameter variable vector; Generate a device state mask vector based on the device's start / stop status. The dimension of the device status mask vector is D; when the device corresponding to the k-th variable in the unified operating parameter variable vector belongs to the current seasonal configuration and is in the on state, ,otherwise =0, This is the device status mask vector corresponding to the k-th variable; Based on the mapping matrix of running parameters and device state mask vector The original running parameter variable vector Mapped to a fixed-dimensional vector of uniform runtime parameter variables The relationship is expressed as:
[0021] Or, represented by elements:
[0022] in, Let j be the j-th original candidate operating parameter variable under the current seasonal configuration during the t-th control period. This refers to the k-th operating parameter variable after unified coding for the t-th control period. The element in the k-th row and j-th column of the runtime parameter mapping matrix. This is the off value corresponding to the kth unified runtime parameter variable, and the symbol ⊙ represents element-wise multiplication; The performance prediction input vector is represented as:
[0023] The low-carbon evaluation input vector is represented as follows:
[0024] in, Let t be the building's heating and cooling load during the t-th control period. Meteorological parameter characteristics, As a time feature, This refers to the characteristics of the equipment's start-up and shutdown status. Characteristics of equipment operating conditions, For regional or time-of-use electricity carbon emission factors, For carbon time series weights, High-carbon period identifier The unified coding of the operating parameter variables for the t-th control period; the input for the performance optimization design of the building energy system consists of the performance prediction input vector and the low-carbon evaluation input vector.
[0025] In one exemplary embodiment of the present invention, the multi-output MLP proxy model includes an input layer, one or more hidden layers, and an output layer. The input of the multi-output MLP proxy model is a building energy system performance prediction input vector, and the output includes the system cooling or heating capacity, the total system power, and the supply and return water temperature difference. The multi-output MLP proxy model is used to simultaneously support system operation energy consumption evaluation, effective COP evaluation, supply and return water temperature difference constraint evaluation, and load satisfaction rate constraint evaluation.
[0026] The multi-output MLP proxy model is represented as:
[0027] In the formula, Let be the input vector for predicting the building energy system performance during the t-th control period. For a set of trainable parameters for a multi-output MLP proxy model, To predict the output vector, For multi-output MLP proxy models The predicted output vector includes:
[0028] in, To predict the cooling or heating load of the system, To predict the total power of the system, To predict the temperature difference between supply and return water; Furthermore, determining the effective energy supply based on the matching relationship between the system's cooling or heating capacity and the building's heating and cooling load includes:
[0029] in, Let t be the building's cooling or heating load during the t-th control period. To provide energy effectively; The effective COP of the system, calculated based on the effective energy supply and total system power, includes:
[0030] in, For the system to effectively COP, To optimize the number of control periods in the time domain, Δt is the control time step. To prevent positive numbers with a denominator of zero.
[0031] The multi-output MLP surrogate model is trained using a multi-output loss function, which includes a system energy supply error term, a system total power error term, and a supply and return water temperature difference error term. This allows the multi-output MLP surrogate model to simultaneously learn the building energy system's energy supply capacity, power consumption, and water system temperature difference response.
[0032] In one exemplary embodiment of the present invention, when using the time-of-use electricity carbon emission factor, carbon time-series weights are constructed based on the ratio of the time-of-use electricity carbon emission factor to the optimized time-domain average carbon emission factor or the normalization result. The carbon time-series weights are determined based on the time-of-use electricity carbon emission factors for each control period within the optimization time domain; the time-of-use electricity carbon emission factor for the t-th control period is denoted as... The average carbon emission factor within the optimized time domain is denoted as The carbon time series weights include ratio-based carbon time series weights and / or normalized carbon time series weights.
[0033]
[0034]
[0035] In the formula, and These represent the maximum and minimum values of the time-sharing electricity carbon emission factor in the optimization time domain, respectively. To prevent positive numbers with a denominator of zero, For carbon time series weights, The normalized carbon time-series weights; Construct high-carbon period identifiers based on carbon time-series weights. When the carbon time-series weight is greater than a preset threshold, the corresponding control period is determined to be a high-carbon period. The carbon time-series weight is used to run at least one of the following: carbon emission objective function, carbon time-series weighted low-carbon scheduling objective, high-carbon period power penalty term, or multi-scenario weight decision rule. The high-carbon period power penalty term is expressed as follows:
[0036] in, This is a power penalty term during high-carbon periods. Optimize the sequence for candidate runtime parameter variables. The total system power predicted by the multi-output MLP surrogate model during the t-th control period is... To control the time step.
[0037] The system calculates operating carbon emissions, carbon emission intensity, and power weighted indexes for high-carbon periods based on total system power, control time step, time-of-use power carbon emission factor, and carbon time series weight.
[0038] In one exemplary embodiment of the present invention, constraint-aware multi-objective optimization uses the optimization sequence of operating parameter variables in the optimization time domain as candidate solutions, and uses system operating energy consumption, operating carbon emission index, system effective COP and fluctuation of operating parameter variables as evaluation indicators; The unified operating parameter variable vector for the t-th control period is represented as:
[0039] in, Let be the j-th operating parameter variable in the t-th control period, and D be the dimension of the fixed-dimensional unified operating parameter variable vector. Let be the set of unified operating parameter variable vectors for the t-th control period; Within the optimization time domain containing T control periods, the complete sequence of optimized runtime parameter variables is represented as follows:
[0040] Where U represents the optimization sequence of candidate running parameter variables in the optimization time domain; The objective function vector of the constraint-aware multi-objective optimization model is represented as:
[0041] in, Let be the objective function. To achieve the system's energy consumption target, In order to achieve carbon emission targets, For the system to achieve effective COP objectives, The objective is to minimize the fluctuation of the operating parameter variables; since a minimization form is used, the effective COP of the system is maximized by minimizing its opposite.
[0042] The system's energy consumption target is expressed as:
[0043] in, To optimize the sequence of candidate runtime parameter variables for multi-output MLP surrogate models The total system power predicted for the t-th control period is as follows. To control the time step; The operational carbon emission targets include operational carbon emissions and / or time-series weighted operational carbon emission targets, where operational carbon emissions are expressed as:
[0044] The carbon time-weighted operating carbon emission index is expressed as:
[0045] in, Let be the regional or time-of-use electricity carbon emission factor for the t-th control period. Let be the carbon time-series weight for the t-th control period; The effective COP objective of the system is expressed as:
[0046] in, For the effective energy supply during the t-th control period, when the system's cooling or heating capacity exceeds the building's cooling or heating load for the corresponding period, the excess supply exceeding the building's cooling or heating load is not included in the effective energy supply. To prevent extremely small positive numbers with a denominator of zero; The target for fluctuation of the running parameter variable is expressed as:
[0047] When t=1, Take the values of the running parameter variables from the actual execution period before the optimization begins or the values of the running parameter variables output from the previous rolling optimization cycle.
[0048] In the multi-objective optimization iteration process, the constraint violation degree of the candidate solution relative to the load satisfaction rate, supply and return water temperature difference, equipment start-up and shutdown logic, and operating parameter engineering constraints is calculated. When the constraint violation degree is zero, the candidate solution is determined to be a feasible candidate solution. When the constraint violation degree is greater than zero, the candidate solution is determined to be a constraint violation candidate solution. At least one of the following is performed: variable pruning, feasible region projection, penalty function penalty, and priority reduction of the external file of non-dominated solution.
[0049] The penalty function is represented as follows:
[0050] in, For the k-th objective function, Let k be the penalty objective function after adding the constraint violation degree. This is the penalty coefficient corresponding to the k-th objective function; Feasible region projection or variable clipping is represented as:
[0051] in, Indicates to feasible region The projection operator, feasible region It is jointly determined by load fulfillment rate constraints, supply and return water temperature difference constraints, equipment start-up and shutdown logic constraints, and engineering executable constraints of operating parameters.
[0052] In one exemplary embodiment of the present invention, calculating the constraint violation degree includes constructing a constraint violation degree function:
[0053]
[0054]
[0055]
[0056]
[0057] In the formula, To constrain the violation degree function, For load satisfaction rate violation degree, and These are the lower and upper limits of the allowable range for load fulfillment rate, respectively. Select the maximum value. Let be the load satisfaction rate for the t-th control period. To prevent the supply and return water temperature difference from violating the standard, Let be the predicted supply and return water temperature difference for the t-th control period. and These are the lower and upper limits of the allowable temperature difference between the supply and return water, respectively. For the degree of out-of-bounds violation of runtime parameters, To optimize the number of control periods in the time domain, and Let $\mathbf$ be the lower and upper limits of the executable parameters of the $j$-th running parameter variable during the $t$-th control period, respectively. For violations of device start / stop logic, Let be the set of operating parameter variables corresponding to the equipment that is in the off state during the t-th control period. This is the off value corresponding to the j-th runtime parameter variable.
[0058] In one exemplary embodiment of the present invention, during constraint-aware multi-objective particle swarm optimization, the candidate solution set and the external archive of non-dominated solutions are updated jointly based on the objective function dominance relationship and constraint violation degree. When both candidate solutions are feasible, they are compared according to the Pareto dominance relationship. When one candidate solution is feasible and the other is a constraint violation candidate, the feasible candidate solution is retained first. When both candidate solutions are constraint violation candidates, the candidate solution with the lower constraint violation degree is retained first. During each iteration or candidate solution update, the multi-output MLP proxy model is used to predict the performance of the candidate operating parameter variable optimization sequence, and simultaneously calculates the objective function and constraint violation degree based on the predicted total system power, effective energy supply, supply and return water temperature difference, time-of-use electricity carbon emission factor and carbon time series weight. When the maximum number of iterations is reached, the update magnitude of the external archive of the non-dominated solution is less than a preset threshold, or the constraint-aware non-dominated solution set satisfies the convergence condition, the Pareto front solution set is output:
[0059] in, For the final Pareto front solution set, For the first There are K candidate optimization design schemes, where K is the number of candidate optimization design schemes.
[0060] One exemplary embodiment of the present invention, determining the preferred solution from the Pareto front solution set, includes: Construct multi-scenario weighted decision rules, which are used to sort the Pareto front solution set obtained by constraint-aware multi-objective optimization. The multi-scenario weighted decision rules include one or more of the following: energy-saving priority scenario, low-carbon priority scenario, efficiency priority scenario, stable operation priority scenario, and equilibrium scenario. Construct evaluation index weight vectors for each scenario, generate decision matrices based on Pareto front solution sets, generate normalized matrices from decision matrices, and obtain weighted normalized matrices through evaluation index weight vectors and normalized matrices.
[0061] In the s-th scenario, construct the evaluation index weight vector:
[0062] in, These are the weights of system operating energy consumption, operating carbon emission indicators, system effective COP, and operating parameter variable fluctuation indicators in the s-th scenario, respectively, and satisfying the following:
[0063] In scenarios prioritizing energy conservation, improve In a low-carbon priority scenario, improve The weights of operational carbon emission indicators are adjusted based on carbon time-series weights, the proportion of high-carbon periods, or power penalties during high-carbon periods; in efficiency-first scenarios, the weights are increased. In scenarios prioritizing stable operation, improve In a balanced scenario, the weights of each evaluation indicator are kept balanced or allocated according to a preset balanced weight. Let the Pareto front solution set be denoted as Construct the decision matrix for the s-th scenario:
[0064] in, Let be the index value of the i-th candidate optimization design scheme under the j-th evaluation index. The evaluation index includes system operating energy consumption, operating carbon emission index, system effective COP and operating parameter variable fluctuation index. The decision matrix is normalized to obtain the normalized matrix:
[0065] in, Let be the normalized value of the i-th candidate solution under the j-th evaluation index. To prevent extremely small positive numbers with a denominator of zero.
[0066] Based on the weighted normalization matrix for the s-th scenario:
[0067] in, Let be the weighted normalized index value for the s-th scenario. Let be the weight of the j-th evaluation metric in the s-th scenario.
[0068] The positive and negative ideal solutions for a certain scenario are determined based on the weighted normalization matrix, and the values of the indicators are selected according to the type of the indicators. Specifically, the positive and negative ideal solutions for the s-th scenario are determined based on the weighted normalization matrix. Among them, the system's effective COP is a benefit-type indicator, while the system's operating energy consumption, operating carbon emission, and operating parameter variable fluctuation indicators are cost-type indicators. For benefit-type indicators, the positive ideal solution takes the maximum value of the indicator, and the negative ideal solution takes the minimum value of the indicator. For cost-type indicators, the positive ideal solution takes the minimum value of the indicator, and the negative ideal solution takes the maximum value of the indicator. The positive and negative ideal solutions for the s-th scenario are represented as follows:
[0069]
[0070] In the formula, For the s-th scenario, For the negative ideal solution in the s-th scenario, Let the value be taken as the positive ideal solution for the m-th and s-th scenarios. Let be the value of the negative ideal solution for the m-th and s-th scenarios.
[0071] Calculate the Euclidean distance between the positive and negative ideal solutions of candidate optimization design schemes in the Pareto front solution set, and calculate their relative closeness in a certain scenario based on the Euclidean distance;
[0072]
[0073] Based on the distances of the i-th candidate solution to the positive and negative ideal solutions, calculate its relative proximity in the s-th scenario:
[0074] In the formula, Let be the Euclidean distance from the i-th candidate optimized design scheme to the s-th positive ideal solution in the scenario. Let be the weighted normalized value of the i-th candidate optimal design scheme under the j-th evaluation index. Let Euclidean distance be the distance from the i-th candidate optimized design scheme to the negative ideal solution in the s-th scenario. and Let be the positive and negative ideal solution values of the j-th evaluation index in the s-th scenario, respectively. The relative approximation of the i-th candidate optimization design scheme in the s-th scenario is given by... To prevent extremely small positive numbers with a denominator of zero.
[0075] All candidate optimization design schemes are ranked according to their proximity, and the candidate optimization design scheme with the highest relative proximity in a certain scenario is selected as the preferred scheme in that scenario.
[0076]
[0077] In the formula, To select the optimal solution, Let i be the i-th candidate solution.
[0078] Output energy-saving and carbon-reduction performance optimization design schemes and hourly operation parameter optimization sequences for building energy systems under scenarios prioritizing energy conservation, low carbon emissions, efficiency, stable operation, and / or balanced operation.
[0079] This invention also provides a computer-aided building energy system performance optimization design data processing system oriented towards time-of-use carbon emission factors, used to execute the above-mentioned computer-aided building energy system performance optimization design data processing method oriented towards time-of-use carbon emission factors, including: Building Energy System Design Data Acquisition Module: Used to acquire building energy system's heating and cooling loads, meteorological parameters, time characteristics, seasonal patterns, equipment configuration, equipment start-up and shutdown status, equipment operating condition characteristics, candidate operating parameter variables, and regional or time-of-use electricity carbon emission factors within a set optimization time domain. Seasonal Equipment Configuration Unified Coding Module: This module generates equipment status masks and operating parameter mapping matrices based on seasonal patterns, equipment configurations, and equipment start-up and shutdown states. It maps different numbers and meanings of operating parameter variables for cooling and heating seasons into a fixed-dimensional unified operating parameter variable vector, and forms a building energy system performance prediction input vector and a low-carbon evaluation input vector. Multi-output MLP proxy model training and performance prediction module: used to train a multi-output MLP proxy model based on simulation model output data or actual operation monitoring data, and predict the system cooling or heating capacity, total system power and supply and return water temperature difference through the multi-output MLP proxy model according to the building energy system performance prediction input vector. Effective Energy Supply and Time-of-Use Carbon Emission Calculation Module: This module is used to determine the effective energy supply based on the matching relationship between the system's cooling or heating capacity and the building's cooling and heating load. It calculates the system's effective COP based on the effective energy supply and the system's total power, and calculates the operating carbon emissions, carbon emission intensity, carbon time series weight, and power weighted index for high-carbon periods based on the system's total power, control time step, and regional or time-of-use electricity carbon emission factors. The constraint violation handling and multi-objective optimization module is used to embed the constraint-aware multi-objective particle swarm optimization algorithm MOPSO with the multi-output MLP surrogate model as a fast performance evaluation function, using the optimization sequence of operating parameter variables as candidate solutions. It calculates the constraint violation degree of the candidate solutions relative to the load satisfaction rate, supply and return water temperature difference, equipment start-up and shutdown logic, and operating parameter engineering constraints. It also performs variable pruning, feasible region projection, penalty function penalty, and / or non-dominated solution external file priority reduction processing on the constraint violation candidate solutions, and generates a Pareto front solution set with system operating energy consumption, operating carbon emission index, system effective COP, and operating parameter variable fluctuation as evaluation indicators. The multi-scenario TOPSIS decision ranking and solution output module is used to perform a comprehensive ranking of the Pareto front solution set by the TOPSIS approximation ideal solution ranking method based on the multi-scenario weight decision rules under the scenarios of energy-saving priority, low-carbon priority, efficiency priority, stable operation priority and / or equilibrium. It outputs the energy-saving and carbon-reduction performance optimization design scheme of the building energy system, the hourly operation parameter optimization sequence, the predicted performance index and the operation carbon emission index under the corresponding scenario.
[0080] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.
[0081] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. This computer software product, stored in a storage medium, includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
[0082] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.
Claims
1. A computer-aided building energy system performance optimization design data processing method for time-sharing carbon emission factors, characterized in that, include: Obtain the target parameters, and after preprocessing some of the target parameters, form a building energy system performance prediction input vector; A multi-output MLP proxy model is constructed and trained based on simulation model output data or actual operation monitoring data. The multi-output MLP proxy model predicts the cooling or heating capacity, total power and supply and return water temperature difference of the system according to the building energy system performance prediction input vector. The effective energy supply is determined based on the matching relationship between the cooling or heating capacity of the system and the building's cooling and heating load, and the effective COP of the system is calculated based on the effective energy supply and the total power of the system. Using the optimization sequence of operating parameter variables as candidate solutions, a constraint-aware multi-objective particle swarm optimization model (MOPSO) is constructed with the optimization objectives of minimizing system operating energy consumption, minimizing operating carbon emission indicators, maximizing system effective COP, and minimizing fluctuations in operating parameter variables. The multi-output MLP proxy model is embedded as a fast performance evaluation function into the constraint-aware multi-objective particle swarm optimization algorithm MOPSO to generate a Pareto front solution set that satisfies the constraints or has been processed for constraint violation. Based on the multi-scenario weight TOPSIS, a compromise optimal solution is determined from the Pareto front solution set, and the energy-saving and carbon-reduction performance optimization design scheme and hourly operation parameter optimization sequence of the building energy system are output. The multi-output MLP proxy model includes an input layer, one or more hidden layers, and an output layer; In the formula, Let be the input vector for predicting the building energy system performance during the t-th control period. For the set of trainable parameters of a multi-output MLP proxy model, To predict the output vector, For multi-output MLP proxy models; The predicted output vector includes: in, To predict the cooling or heating load of the system, To predict the total power of the system, To predict the temperature difference between supply and return water; The determination of effective energy supply based on the matching relationship between the system's cooling or heating capacity and the building's cooling and heating load includes: in, Let t be the building's cooling or heating load during the t-th control period. To provide energy effectively; The effective COP of the system is calculated based on the effective energy supply and total system power, including: in, For the system to effectively COP, To optimize the number of control periods in the time domain, Δt is the control time step. To prevent positive numbers with a denominator of zero.
2. The data processing method for computer-aided building energy system performance optimization design based on time-of-use carbon emission factors as described in claim 1, characterized in that, Also includes: When using time-of-use electricity carbon emission factors, carbon time-series weights are constructed based on the ratio of time-of-use electricity carbon emission factors to the optimized time-domain average carbon emission factors or the normalization result. In the formula, and These represent the maximum and minimum values of the time-sharing electricity carbon emission factor in the optimization time domain, respectively. To prevent positive numbers with a denominator of zero, For carbon time series weights, For normalized carbon time series weights, To optimize the average carbon emission factor in the time domain, Let be the time-of-use electricity carbon emission factor for the t-th control period; The system calculates operating carbon emissions, carbon emission intensity, and power weighted indexes for high-carbon periods based on total system power, control time step, time-of-use power carbon emission factor, and carbon time series weight.
3. The data processing method for computer-aided building energy system performance optimization design based on time-of-use carbon emission factors as described in claim 1, characterized in that, The constraint-aware multi-objective particle swarm optimization model includes: in, Let be the objective function. To achieve the system's energy consumption target, In order to achieve carbon emission targets, For the system to achieve effective COP objectives, The target is the fluctuation of the running parameter variables; During the multi-objective optimization iteration process, the constraint violation degree of the candidate solution relative to the load satisfaction rate, supply and return water temperature difference, equipment start-up and shutdown logic, and operating parameter engineering constraints is calculated. When the constraint violation degree is equal to zero, the candidate solution is determined as a feasible candidate solution; when the constraint violation degree is greater than zero, the candidate solution is determined as a constraint violation candidate solution, and at least one of the following is performed: variable pruning, feasible region projection, penalty function penalty, and non-dominated solution external file priority reduction.
4. The data processing method for computer-aided building energy system performance optimization design based on time-of-use carbon emission factors as described in claim 3, characterized in that, Calculating the constraint violation degree includes constructing a constraint violation degree function: In the formula, To constrain the violation degree function, For load satisfaction rate violation degree, and These are the lower and upper limits of the allowable range for load fulfillment rate, respectively. Select the maximum value. For the first Load fulfillment rate for each control period To prevent the supply and return water temperature difference from violating the standard, Let be the predicted supply and return water temperature difference for the t-th control period. and These are the lower and upper limits of the allowable temperature difference between the supply and return water, respectively. For the degree of out-of-bounds violation of runtime parameters, To optimize the number of control periods in the time domain, To unify the dimensions of the runtime parameter variable vector to a fixed dimension, and The first The number of runtime parameter variables in the first... The lower and upper limits of the executable capacity of the project during each control period. For the first The first control period One runtime parameter variable, For violations of device start / stop logic, For the first The set of operating parameter variables corresponding to the equipment that is in the off state during each control period. For the first The corresponding shutdown value for each running parameter variable.
5. The data processing method for computer-aided building energy system performance optimization design based on time-of-use carbon emission factors according to claim 3, characterized in that, Also includes: In the constraint-aware multi-objective particle swarm optimization process, the candidate solution set and the external archive of non-dominated solutions are updated together based on the dominance relationship of the objective function and the degree of constraint violation. When both candidate solutions are feasible, they are compared according to the Pareto dominance relation; when one candidate solution is feasible and the other is a constraint violation candidate solution, the feasible candidate solution is preferred; when both candidate solutions are constraint violation candidates, the candidate solution with the smaller constraint violation degree is preferred. When the maximum number of iterations is reached or the constraint-aware non-dominated solution set satisfies the convergence condition, the Pareto front solution set is output.
6. The data processing method for computer-aided building energy system performance optimization design based on time-of-use carbon emission factors as described in claim 1, characterized in that, The compromise optimal solution determined from the Pareto front solution set based on multi-scenario weighted TOPSIS includes: Construct multi-scenario weighted decision rules, which include one or more of the following: energy-saving priority scenario, low-carbon priority scenario, efficiency priority scenario, stable operation priority scenario, and balanced scenario; Construct evaluation index weight vectors for each scenario, generate decision matrices based on Pareto front solution sets, generate normalized matrices from decision matrices, and obtain weighted normalized matrices through evaluation index weight vectors and normalized matrices.
7. The data processing method for computer-aided building energy system performance optimization design based on time-of-use carbon emission factors as described in claim 6, characterized in that, Also includes: The positive and negative ideal solutions for a certain scenario are determined based on the weighted normalization matrix, and the values of the indicators are selected according to the type of the indicators. Calculate the Euclidean distance between the positive and negative ideal solutions of candidate optimization design schemes in the Pareto front solution set, and calculate their relative closeness in a certain scenario based on the Euclidean distance; All candidate optimization design schemes are sorted according to the proximity, and the candidate optimization design scheme with the highest relative proximity in a certain scenario is selected as the preferred scheme in that scenario.
8. A computer-aided building energy system performance optimization design data processing system based on time-of-use carbon emission factors, characterized in that, A data processing method for performing computer-aided building energy system performance optimization design based on time-of-use carbon emission factors as described in any one of claims 1-7, comprising: The building energy system design data acquisition module is configured to acquire target parameters and preprocess some of the target parameters to form a building energy system performance prediction input vector. Seasonal Equipment Configuration Unified Coding Module: This module generates equipment status masks and operating parameter mapping matrices based on seasonal patterns, equipment configurations, and equipment start-up and shutdown states. It maps different numbers and meanings of operating parameter variables for cooling and heating seasons into a fixed-dimensional unified operating parameter variable vector, and forms a building energy system performance prediction input vector and a low-carbon evaluation input vector. The multi-output MLP proxy model training and performance prediction module is configured to build and train a multi-output MLP proxy model based on simulation model output data or actual operation monitoring data. The multi-output MLP proxy model predicts the cooling or heating capacity, total power and supply and return water temperature difference of the system according to the building energy system performance prediction input vector. The effective energy supply and time-of-use carbon emission index calculation module is configured to determine the effective energy supply based on the matching relationship between the system's cooling or heating capacity and the building's cooling and heating load, and to calculate the system's effective COP based on the effective energy supply and the system's total power. The constraint violation handling and multi-objective optimization module is configured to use the optimization sequence of operating parameter variables as candidate solutions to construct a constraint-aware multi-objective particle swarm optimization (MOPSO) model with the optimization objectives of minimizing system operating energy consumption, minimizing operating carbon emission indicators, maximizing system effective COP, and minimizing fluctuations of operating parameter variables. The multi-scenario TOPSIS decision ranking and solution output module is configured to embed the multi-output MLP proxy model as a fast performance evaluation function into the constraint-aware multi-objective particle swarm optimization algorithm MOPSO, generate a Pareto front solution set that satisfies the constraints or has been processed for constraint violation, determine the compromise optimal solution from the Pareto front solution set based on the multi-scenario weighted TOPSIS, and output the energy-saving and carbon-reduction performance optimization design scheme of the building energy system and the hourly operation parameter optimization sequence.
Citation Information
Patent Citations
Building carbon emission prediction and control method based on machine learning and multi-objective optimization
CN120317458A
Building carbon emission multi-objective optimization method and system based on proxy model
CN122197334A
